Neural Network Inverse Control Of Ship Rudder/Fin Joint Roll Stabilization System

Reducing ship roll motion is one of the important tasks of ship motion control.Rudder/fin joint control system is a comprehensive roll stabilization device developed in fin stabilizer and steering control system,which can achieve good performance on roll stabilization and ensure the accuracy of course control.This thesis focuses on the study about establishing the rudder/fin joint system nonlinear mathematical model,and the design of rudder/fin joint roll stabilization neural network inverse control system.Firstly,based on the separated modeling theory,rudder/fin joint system nonlinear mathematical model of 4 degrees of freedom(surge,sway,roll and yaw)is established,which is affected by wind and waves disturbance.Besides,ship's maneuverability simulation test is carried out to verify the correctness of the model built.Then,the inverse control method which can realize linear decoupling is combined with the neural network with learning and function approximation ability to study the inverse neural network design of ship fin stabilizer system and rudder/fin joint system.(1)The invertibility of the nonlinear mathematical model of the ship fin stabilizer system is analyzed.The inverse system of the fin stabilizer neural network is constructed.A linear controller is designed for close loop control for the pseudo-linear composite system composed of the inverse system and the original system.(2)The invertibility of the rudder/fin joint system is analyzed.Based on the inverse control principle of neural network,the pseudo-linear composite system of the rudder/fin joint system is constructed.The linear PID controller is designed to realize the neural network inverse control of the rudder/fin joint system.(3)Aiming at the difficulty of adjusting the parameters of linear PID controller for the above-mentioned rudder/fin roll stabilization neural network inverse control system,a PID controller based on BP neural network tuning is designed to realize the self-tuning of controller parameters.(4)Aiming at the inverse modeling error of neural network in the process of constructing pseudo-linear composite system,a neural network inverse model/PID composite controller is designed.Based on the open-loop inverse control of BP neural network,the BP neural network direct inverse control system with PID compensation is proposed to overcome the inverse modeling error and achieve effective control of the controlled object.Finally,aiming at the above neural network inverse control and its improved algorithm,the rudder/fin roll stabilization simulations are performed under different sea conditions.The simulation results verify the effectiveness of the designed controller and achieve the purpose of controller optimization,which can enable the ship to achieve a better roll stabilization effect while ensuring the control accuracy of course control.